As reported in https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=906609 GnuCash FTBFS on mips due to segfault related to endianness. Debian bug report contains extensive analysis by Bernhard Übelacker and proposed Debian-specific patch based on passing `-DWORDS_BIGENDIAN` and the following: ~~~~ --- a/common/config.h.cmake.in +++ b/common/config.h.cmake.in @@ -355,9 +355,9 @@ # define WORDS_BIGENDIAN 1 # endif #else # ifndef WORDS_BIGENDIAN -#cmakedefine WORDS_BIGENDIAN +#cmakedefine WORDS_BIGENDIAN 1 # endif #endif /* Define to 1 if on MINIX. */ ~~~~ That seems to help (though perhaps needs to be incorporated into CMake somehow) but then "test-backend-dbi" failed on mips: https://buildd.debian.org/status/fetch.php?pkg=gnucash&arch=mips&ver=1%3A3.3-2&stamp=1538887923&raw=0
We got halfway there with https://github.com/Gnucash/gnucash/commit/766dc9b2d5cee4810a2d3a3b85091e1934f3d19f#diff-af3b638bc2a3e6c650974192a53c7291 but left off the config.h part.
I've pushed the fix to config.h.cmake.in. I'm not sure what to make of the failure in test-backend-dbi. The warning is nonsensical: (/<<PKGBUILDDIR>>/.build/bin/test-backend-dbi:31612): GLib-GObject-WARNING **: 04:46:36.835: value "0" of type 'gint' is invalid or out of range for property 'fraction' of type 'gint' How can 0 be invalid or out of range for int? It's the one number that reads the same for both big and little endianness. Finding what's actually wrong will require someone to go at it with a debugger, I'm afraid.
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=906609 is resolved so I guess this is too.
